TheySaid Review

Real user testing is slow and expensive, so teams often ship without it. TheySaid is a platform built to speed that up with AI, using synthetic testers, AI stand-in users trained on your customer data, alongside real human audiences.

It runs AI surveys, interviews, forms and polls, offers an expert AI reviewer and a spec check, works in 70+ languages, and plugs into developer tools via MCP and support-ticket triggers. As a research tool it sits near platforms like Survicate, but its defining idea is the synthetic tester.

This review covers how TheySaid tests with AI, which teams it suits, how its plans work, and the honesty point at its core.

TheySaid Synthetic-Testing Facts

TheySaid, The AI-User-Testing Brief

• TheySaid is AI user testing: synthetic testers trained on your data, plus human audiences.
• It runs research formats: AI surveys, interviews, forms and polls in 70+ languages.
• It fits dev workflows: an MCP server and testing triggered from support tickets.
• Read results honestly: synthetic findings come from AI-simulated users, not real humans.

AI User Testing With Synthetic And Human Testers

TheySaid's purpose is to make user research fast and cheap by simulating it. Its headline feature is synthetic testers, AI stand-in users trained on your own customer data, that can respond to a design, flow or question as a plausible user might.

Around that sit the familiar research formats, AI surveys, interviews, forms and polls, plus voice conversations, an expert AI reviewer and a spec check, across 70+ languages, with clips and highlight reels to share findings.

It is built to fit modern workflows: an MCP server lets it run from tools like Claude, Cursor and Replit, and testing can be triggered automatically when a support ticket moves, so research happens continuously rather than as a one-off study.

The honest read, and the key framing, is that TheySaid's core output is AI-simulated, not real human feedback, so synthetic-test findings must never be presented as if real users said them. To its credit the vendor is fairly transparent, publishing a human-versus-synthetic agreement figure and offering real human panels; still, its own comparison table is unpopulated (placeholder zeros), and one testimonial shows mismatched paired numbers, so read the marketing carefully.

TheySaid makes research fast with synthetic testers, which is genuinely useful, but their output is AI-simulated, not real human feedback, and must not be presented as such.

How Does TheySaid Test With AI?

TheySaid pairs synthetic testing with research formats and workflow hooks.

Synthetic Testers Trained On Your Data

Synthetic testers, trained on your customer data, respond to designs and questions, while human audiences remain available for when you need real people, and an expert AI reviewer and spec check add structure.

Where a survey tool like Survicate collects responses from real users, TheySaid can simulate a first pass with synthetic testers before (or instead of) recruiting humans.

Surveys, Interviews And A Spec Check

AI surveys, interviews, forms, polls and voice conversations run in 70+ languages, with clips and highlight reels, and MCP plus ticket triggers let testing run inside development and support workflows.

For structured survey-based market research specifically, a tool such as quantilope is a different, human-data approach, and a synthetic-research peer like Synthetic Users is worth weighing; TheySaid's distinctive angle is the synthetic tester and the continuous, workflow-triggered testing.

Which UX Teams TheySaid Fits

TheySaid suits product, design and research teams that want faster, cheaper validation and understand the difference between synthetic and real feedback.

It suits teams that want research to run continuously inside their tools (via MCP and ticket triggers), and those that value having both synthetic and human audiences in one platform.

It fits less well teams whose decisions demand real human evidence, and anyone who would present synthetic findings as genuine user research to stakeholders.

A fair way to judge TheySaid is by how you would use its output. As a fast, cheap first pass, catching obvious issues, pressure-testing a flow, sizing a question before you spend on real users, synthetic testing is genuinely valuable and the workflow hooks make it effortless. As a replacement for real human evidence on a high-stakes decision, it is the wrong tool, so the honest practice is to treat synthetic results as a hypothesis and confirm the important ones with the human panels it also offers.

How TheySaid Opens Free And Paid Testing

TheySaid is freemium, free to start with no credit card, with usage-based pricing (per synthetic test) and an enterprise tier via demo. We do not publish prices, which change; check the site for current tiers.

The value question is whether faster, cheaper synthetic validation saves enough time and research spend to justify the cost, for how much testing you do and how much you can act on simulated findings.

The sensible approach is to use the free entry to run a real synthetic test on an actual design or question, compare its findings against what you already know (or a small human check), and judge how much to trust it before scaling. Keep synthetic and human clearly labelled in anything you share, and lean on the human panels for decisions that need real evidence.

Running One Real Synthetic Test In TheySaid

Because synthetic is not human, trial TheySaid on a real question you can check.

1. Run a synthetic test on a real design or question.
2. Compare its findings against what you already know.
3. Run a small human check on the same question.
4. See how closely synthetic and human agree for you.
5. Try the MCP or ticket trigger if it fits your workflow.
6. Label synthetic vs human clearly in anything you share.
7. Use human panels for high-stakes decisions.